The research of color image segmentation attracts more and more people's attention for color image has more information than gray image. In this paper, a novel color image segmentation method is proposed based on color similarity in RGB color space. According to the color and luminance information in RGB color space, the dominant color is determined firstly; then a scale invariant and semantic mathematic model, called SIMILATION, is exploited to calculating color similarity, and Combined with the proposed calculations method of color component, which creates a color-class map. Finally, the information of the corresponding color-class map is utilized to classify the pixels. Experimental results clearly show that the proposed color similarity based color image segmentation algorithm is accurate, robust with low computational complexity.